Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:125075
Feedback-ID: ifab94697:Fastmail
Precedence: bulk
MIME-Version: 1.0
Date: Tue, 20 Aug 2024 23:53:48 +0200
To: internals@lists.php.net
Message-ID: <a9647db8-2330-4834-a5ee-5047c09edee7@app.fastmail.com>
In-Reply-To: <00ee5318-329c-4a85-a2cd-1fc7050103ba@app.fastmail.com>
References: <2716f729-4008-4f75-8412-861d8960b746@app.fastmail.com>
 <cOoRTbvWeeI8DDj531R9b7NwiyLusi5oHFI_0FmZeOx-R5bs1dUC7VO5ZypChYBECwA8SBzrpEABGGH0mPKrLDsYMSfrhuZX2Gn6zUyP4fI=@gpb.moe>
 <26338153-6d16-456a-81bf-8231bdaf1b79@app.fastmail.com>
 <6197f6e1-d123-41e1-87c8-887c6508bfa2@rwec.co.uk>
 <97d993ee-c558-43a1-a36a-8e06aa2d829a@app.fastmail.com>
 <1E9FCFA6-4FAD-4F47-9FC6-6AA8F2FD6BBF@rwec.co.uk>
 <00ee5318-329c-4a85-a2cd-1fc7050103ba@app.fastmail.com>
Subject: Re: [PHP-DEV] function autoloading v4 RFC
Content-Type: multipart/alternative;
 boundary=4bf74522a7ae44dc89a5800a4a1fdaac
From: rob@bottled.codes ("Rob Landers")

--4bf74522a7ae44dc89a5800a4a1fdaac
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable



On Tue, Aug 20, 2024, at 18:07, Rob Landers wrote:
> On Tue, Aug 20, 2024, at 08:50, Rowan Tommins [IMSoP] wrote:
>>=20
>>=20
>> On 20 August 2024 00:21:22 BST, Rob Landers <rob@bottled.codes> wrote:
>> >
>> >I assume you are worried about something like this passing test?
>> >
>> >--TEST--
>> >show called only once
>> >--FILE--
>> ><?php
>> >
>> >namespace test;
>> >
>> >spl_autoload_register(function($name) {
>> >    echo "name=3D$name\n";
>> >}, true, false, SPL_AUTOLOAD_FUNCTION);
>> >
>> >echo strlen('foo');
>> >echo strlen('bar');
>> >echo strlen('baz');
>> >?>
>> >--EXPECT--
>> >name=3Dtest\strlen
>> >333
>> >
>> >In my RFC, I mention it is called exactly once.
>>=20
>>=20
>> I haven't looked at the PR, only the RFC, and I did not see this very=
 important detail explained anywhere. The only thing I can see is this r=
ather ambiguous sentence:
>>=20
>> >  The function autoloader will not be called again.=20
>>=20
>> That could mean not called again for the current call (compared with =
proposals that call it a second time with the unequalled name); it could=
 mean not called again for the current line of code (based on the curren=
t caching behaviour); or never called again for that combination of name=
space and name; or possibly, never called again for that combination of =
namespace, name, and callback function.
>>=20
>> That's not a small detail of the implementation, it's a really fundam=
ental difference from previous proposals.=20
>>=20
>> So I would like to repeat my first response to your RFC: that it shou=
ld sound more time explaining your approach to the multiple lookup probl=
em.
>>=20
>> Regards,
>> Rowan Tommins
>> [IMSoP]
>>=20
>=20
> Thanks Rowan,
>=20
> That's a fair critique.
>=20
> I expect some of the wording will be more clear once I write out the d=
ocumentation -- even if it isn't used directly, I tend to write out docu=
mentation to force myself to reconcile the code with the plan, find logi=
c bugs, perform larger scale tests, and create tests to verify assertion=
s in the documentation. From there, I'll update the plan or code to get =
everything to match and spend some time on clarity. It's the hardest par=
t, IMHO, as it requires diligently ensuring everything is correct. In ot=
her words, writing the documentation makes it feel like a "real thing" a=
nd it triggers what small amount of perfectionism I have.
>=20
> =E2=80=94 Rob

I have an experimental library that I use for testing these kinds of thi=
ngs. There are aspects of it that I could work with to make use of funct=
ion autoloading. Thus, I did so and benchmarked the performance of unit =
tests. The unit testing library makes a ton of "unqualified function cal=
ls".

I spent some time working on two autoloaders:
 1. A naive autoloader: parses out the file to load, checks if it exists=
, and then requires the file.
 2. An optimized autoloader: only cares about the namespace it has regis=
tered. All others are an instant return.
In the "vanilla" case, I was mostly concerned with variation. I wanted a=
 statistically significant result, so once I got my system into a stable=
 state and I was no longer seeing any variance, I started benchmarking.=20

For the naive autoloader, I saw a performance degradation of about 6% an=
d lots of variability. This is probably due to the "file_exists" check b=
eing done every time an unqualified name was called.

However, for the optimized autoloader, I ended up with less variability =
(=F0=9F=A4=94) than the vanilla approach and absolutely no measurable pe=
rformance degradation.

Now, time to try this on a larger scale... WordPress. It's pretty much t=
he only large codebase I know of that makes use of tons of functions.

=E2=80=94 Rob
--4bf74522a7ae44dc89a5800a4a1fdaac
Content-Type: text/html; charset=utf-8
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE html><html><head><title></title><style type=3D"text/css">=0Ap.=
MsoNormal,p.MsoNoSpacing{margin:0}=0Ap.MsoNormal,p.MsoNoSpacing{margin:0}=
=0Ap.MsoNormal,p.MsoNoSpacing{margin:0}</style></head><body><div><br></d=
iv><div><br></div><div>On Tue, Aug 20, 2024, at 18:07, Rob Landers wrote=
:<br></div><blockquote type=3D"cite" id=3D"qt" style=3D""><div>On Tue, A=
ug 20, 2024, at 08:50, Rowan Tommins [IMSoP] wrote:<br></div><blockquote=
 type=3D"cite" id=3D"qt-qt" style=3D""><div><br></div><div><br></div><di=
v>On 20 August 2024 00:21:22 BST, Rob Landers &lt;<a href=3D"mailto:rob@=
bottled.codes">rob@bottled.codes</a>&gt; wrote:<br></div><div>&gt;<br></=
div><div>&gt;I assume you are worried about something like this passing =
test?<br></div><div>&gt;<br></div><div>&gt;--TEST--<br></div><div>&gt;sh=
ow called only once<br></div><div>&gt;--FILE--<br></div><div>&gt;&lt;?ph=
p<br></div><div>&gt;<br></div><div>&gt;namespace test;<br></div><div>&gt=
;<br></div><div>&gt;spl_autoload_register(function($name) {<br></div><di=
v>&gt;&nbsp;&nbsp;&nbsp; echo "name=3D$name\n";<br></div><div>&gt;}, tru=
e, false, SPL_AUTOLOAD_FUNCTION);<br></div><div>&gt;<br></div><div>&gt;e=
cho strlen('foo');<br></div><div>&gt;echo strlen('bar');<br></div><div>&=
gt;echo strlen('baz');<br></div><div>&gt;?&gt;<br></div><div>&gt;--EXPEC=
T--<br></div><div>&gt;name=3Dtest\strlen<br></div><div>&gt;333<br></div>=
<div>&gt;<br></div><div>&gt;In my RFC, I mention it is called exactly on=
ce.<br></div><div><br></div><div><br></div><div>I haven't looked at the =
PR, only the RFC, and I did not see this very important detail explained=
 anywhere. The only thing I can see is this rather ambiguous sentence:<b=
r></div><div><br></div><div>&gt;&nbsp; The function autoloader will not =
be called again.&nbsp;<br></div><div><br></div><div>That could mean not =
called again for the current call (compared with proposals that call it =
a second time with the unequalled name); it could mean not called again =
for the current line of code (based on the current caching behaviour); o=
r never called again for that combination of namespace and name; or poss=
ibly, never called again for that combination of namespace, name, and ca=
llback function.<br></div><div><br></div><div>That's not a small detail =
of the implementation, it's a really fundamental difference from previou=
s proposals.&nbsp;<br></div><div><br></div><div>So I would like to repea=
t my first response to your RFC: that it should sound more time explaini=
ng your approach to the multiple lookup problem.<br></div><div><br></div=
><div>Regards,<br></div><div>Rowan Tommins<br></div><div>[IMSoP]<br></di=
v><div><br></div></blockquote><div><br></div><div>Thanks Rowan,<br></div=
><div><br></div><div>That's a fair critique.<br></div><div><br></div><di=
v>I expect some of the wording will be more clear once I write out the d=
ocumentation -- even if it isn't used directly, I tend to write out docu=
mentation to force myself to reconcile the code with the plan, find logi=
c bugs, perform larger scale tests, and create tests to verify assertion=
s in the documentation. From there, I'll update the plan or code to get =
everything to match and spend some time on clarity. It's the hardest par=
t, IMHO, as it requires diligently ensuring everything is correct. In ot=
her words, writing the documentation makes it feel like a "real thing" a=
nd it triggers what small amount of perfectionism I have.<br></div><div>=
<br></div><div id=3D"qt-sig121229152">=E2=80=94 Rob<br></div></blockquot=
e><div><br></div><div>I have an experimental library that I use for test=
ing these kinds of things. There are aspects of it that I could work wit=
h to make use of function autoloading. Thus, I did so and benchmarked th=
e performance of unit tests. The unit testing library makes a ton of "un=
qualified function calls".<br></div><div><br></div><div> I spent some ti=
me working on two autoloaders:<br></div><ol><li>A naive autoloader: pars=
es out the file to load, checks if it exists, and then requires the file=
.<br></li><li>An optimized autoloader: only cares about the namespace it=
 has registered. All others are an instant return.<br></li></ol><div>In =
the "vanilla" case, I was mostly concerned with variation. I wanted a st=
atistically significant result, so once I got my system into a stable st=
ate and I was no longer seeing any variance, I started benchmarking. <br=
></div><div><br></div><div>For the naive autoloader, I saw a performance=
 degradation of about 6% and lots of variability. This is probably due t=
o the "file_exists" check being done every time an unqualified name was =
called.<br></div><div><br></div><div>However, for the optimized autoload=
er, I ended up with less variability (=F0=9F=A4=94) than the vanilla app=
roach and absolutely no measurable performance degradation.<br></div><di=
v><br></div><div>Now, time to try this on a larger scale... WordPress. I=
t's pretty much the only large codebase I know of that makes use of tons=
 of functions.</div><div><br></div><div id=3D"sig121229152">=E2=80=94 Ro=
b<br></div></body></html>
--4bf74522a7ae44dc89a5800a4a1fdaac--